Re: [HTCondor-users] Networking of the Worker-nodes in Flocks

[Ben Cotton <ben.cotton@xxxxxxxxxxxxxxxxxx>]

Salman,

After a match is made, the scheduler and execute node must be able to
communicate with each other in order to start and run the job. If the
clusters are on private networks, you can use HTCondor Connection
Brokering (CCB) to allow the nodes to communicate. A common setup is
to have the central manager (and potentially the schedulers) on public
IP space and have the execute nodes on private IP space. Section 3.7.4
of the HTCondor Manual describes CCB:

http://research.cs.wisc.edu/htcondor/manual/v8.2/3_7Networking_includes.html#SECTION00474000000000000000

Slide 4 of Alan DeSmet's talk from this year's HTCondor Week gives an
idea of the communication paths between daemons:

http://research.cs.wisc.edu/htcondor/HTCondorWeek2014/presentations/DeSmetA_AdminTutorial.pdf


Thanks,
BC

-- 
Ben Cotton
main: 888.292.5320

Cycle Computing
Leader in Utility HPC Software

http://www.cyclecomputing.com
twitter: @cyclecomputing